janwilmake/efficient-recorder: Attempt to create an Open Source Privacy Focused Rewind.ai Alternative for data capture

Today I was inspired by this tweet after which I fell into this rabbit hole to create an efficient recorder for your Screen, System audio, and Mic.

After trying to implement this using Claude in Swift I completely failed (see folder swift-version) so I decided to create a simpler version (also using Claude) in Node.js

An intelligent, multi-modal recording CLI tool that automatically captures and streams audio, screenshots, and webcam video to S3. Uses advanced detection and efficient resource management to capture multimedia content.

• Intelligent Audio Recording:

  • Monitors audio at 8kHz with automatic speech detection
  • Switches to high-quality 44.1kHz recording when speech is detected
  • Automatically stops recording after 2 seconds of silence
  • Direct streaming to S3

• Automated Screenshot Capture:

  • Configurable screenshot interval
  • Immediate upload of screenshots to S3
  • Low-overhead screen capture

• Webcam Video Capture:

  • Configurable webcam capture interval
  • Direct upload of webcam images to S3
  • Supports custom webcam device selection

• Efficient Resource Management:

  • Minimal system resource usage during idle periods
  • Intelligent detection and recording mechanisms
  • Concurrent upload processing

This package requires:

You’ll need:

• S3-compatible storage endpoint
• Access key
• Secret key
• A bucket named “recordings” (or modify the code to use a different bucket name)

Run the recorder using npx:

npx efficient-recorder \
  --endpoint YOUR_S3_ENDPOINT \
  --key YOUR_ACCESS_KEY \
  --secret YOUR_SECRET_KEY \
  --enable-screenshot \
  --screenshot-interval 5000 \
  --enable-webcam \
  --webcam-interval 3000 \
  --image-quality 80

• --endpoint: Your S3-compatible storage endpoint URL
• --key: Your AWS/S3 access key
• --secret: Your AWS/S3 secret key
• --enable-screenshot: Enable screenshot capture
• --screenshot-interval: Interval between screenshots (ms)
• --enable-webcam: Enable webcam capture
• --webcam-interval: Interval between webcam captures (ms)
• --webcam-device: Specify webcam device (optional)
• --image-quality: Image quality for webcam/screenshots (1-100)

1. Audio Monitoring

   • Continuous low-quality audio monitoring
   • Switches to high-quality recording when speech is detected
   • Automatic recording start and stop based on sound levels

2. Screenshot Capture

   • Captures screenshots at specified intervals
   • Immediate upload to S3
   • Configurable capture frequency

3. Webcam Capture

   • Captures webcam images at specified intervals
   • Supports multiple webcam devices
   • Immediate upload to S3

4. Efficient Upload

   • Queued upload processing
   • Concurrent uploads with multi-part support
   • Minimal system resource overhead

Audio

• Monitoring: 8kHz, Mono
• Recording: 44.1kHz, Stereo
• Format: WAV (16-bit PCM)

Screenshots

• Captured at system screen resolution
• Uploaded as PNG

Webcam

• Resolution: 1280×720
• Format: JPEG
• Configurable quality

1. “Command not found: rec”

   • Ensure SoX is installed correctly
   • Verify SoX is in your system PATH

2. S3 Upload Issues

   • Check S3 credentials
   • Verify bucket exists and write permissions are granted

3. No Audio/Video Input

   • Check system input devices
   • Verify microphone and webcam permissions

MIT License – See LICENSE file for details

Contributions are welcome! Please submit a Pull Request.

For issues and feature requests, please open an issue on GitHub.